Kin Recognition by Phenotype Matching

Abstract

In this paper we developed models to explore possible mechanisms by which individuals could classify conspecifics as full siblings, half siblings, or nonrelatives, assuming that kinship is assessed by comparing phenotypes of conspecifics to a learned template. The hypothesized template is learned during interactions with one's parents or one's self. In general, phenotype matching based on fully heritable metric traits was more accurate than discriminations based on discretely varying traits (table 1). However, in the metric trait models recognition errors rapidly increased if random environmental factors influenced the observed variation. Reliable discrimination of kinship based on discrete traits required either a large number of variants (alleles per locus), or the assessment of many traits (independently assorting loci). For a fixed total number of alleles, over all recognition loci, minimum error was obtained with three to five alleles per locus. Discrimination between full siblings and either half sibs or nonrelatives based on comparisons to one's self was less accurate when completely dominant traits were assessed than when there was codominant expression of the alleles. However, dominance increased the accuracy of discrimination between half sibs and non-relatives based on trait matching to one's self. Dominance also increased the accuracy of discrimination among all three categories of relatives when phenotype comparisons were made to one's parents. Phenotype matching to one's self allows discrimination of full sibs from nonrelatives with no more, and often less, error than does comparison to either parent. The best source of information for discriminating half sibs from either full sibs or nonrelatives is often the individual that is a parent of only one of the two kin classes being differentiated. Many attributes, metric or discrete, may not possess the variation necessary for reliable recognition of kin. However, selection might favor increased genetically determined variation in those characters specifically involved in phenotype matching.